CN104933715A - Registration method applied to retina fundus image - Google Patents
Registration method applied to retina fundus image Download PDFInfo
- Publication number
- CN104933715A CN104933715A CN201510333654.0A CN201510333654A CN104933715A CN 104933715 A CN104933715 A CN 104933715A CN 201510333654 A CN201510333654 A CN 201510333654A CN 104933715 A CN104933715 A CN 104933715A
- Authority
- CN
- China
- Prior art keywords
- registration
- feature
- image
- point
- images
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
- G06T7/0014—Biomedical image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30041—Eye; Retina; Ophthalmic
Abstract
The invention discloses a registration method applied to a retina fundus image, and belongs to the field of medical image processing. On the basis of an SIFT (Scale Invariant Feature Transform) method, a new feature point detection and feature description method is proposed, the detection amount and the registration precision of the feature points of the retina fundus image can be improved, and an image registration effect is better. In order to further improve registration accuracy, a mismatching fiducial point rejection strategy is introduced into registration. A great quantity of registration experiments indicates that the registration method can realize the accurate and stable registration of the retina fundus image and overcomes a defect that the SIFT method can not realize the registration in parts of retina fundus images.
Description
Technical field
The invention belongs to field of medical image processing, be specifically related to a kind of method for registering being applied to retinal fundus images.
Background technology
Retinal fundus images is the objective diagnosis image of a kind of standard that oculist commonly uses, and it is obtained by fundus camera.Eyeground refers to the general designation of interior eye rear portion institutional framework, and it generally comprises retina, optic nerve, choroid and macula lutea etc.Research shows, the diseases such as glaucoma, DRP and coronary heart disease can cause visual impairment, are the main causes causing blind.In the process of clinic diagnosis, oculist needs to be analyzed the retinal fundus images of same patient's different times shooting.Due to the difference of imaging angle and imaging circumstances, at this time need to carry out high-precision registration to two width images.In addition, due to the viewing angle problem of fundus camera, being limited in scope of every width shooting, just needs multiple image to be carried out fusion and whether occurs lesion region to observe eye.Therefore, need to carry out registration to these retinal fundus images, prepare for the fusion of follow-up retinal fundus images and eyeground pathological changes detect.Generally speaking, the diagnosis of retinal fundus images registration to PVR serves vital role, so the Study of Registration of retinal fundus images has great significance to the early detection of fundus oculi disease, auxiliary diagnosis.
Based on SIFT (Scale Invariant Feature Transform) method as one of current method for registering the most concerned in the registration of medical image, it is very stable that it extracts unique point of image, also solves the registration problems occurred between image in the situations such as translation, rotation, affined transformation and light change simultaneously.But very important is a bit, the feature that the method is extracted is not the angle point in people's vision meaning, can a large amount of unique points be detected in the medical image that some contrast is high, will redundant information be produced like this, cause and can have a strong impact on efficiency when feature interpretation; And in the medical image that contrast is lower, as in retinal fundus images, because SIFT method for registering is comparatively strict to the testing conditions of unique point candidate point, the unique point quantity that parts of images detects very little, affects the precision of registration.
Summary of the invention
For the registration fault that the SIFT method of classics cannot realize at part retinal fundus images, the present invention devises a kind of method for registering being applied to eye ground image.The present invention introduces new feature point detection and character description method in registration process, improves eye ground image characteristic point amount detection and registration accuracy.In order to further improve the degree of accuracy of registration, in registration, introducing Mismatching point reject strategy.Core technology scheme of the present invention mainly contains following two aspects:
1. introduce new feature point detecting method and character description method
As shown in Figure 1, a feature structure comprises three neighbor nodes of main bifurcation and its periphery, and wherein, the length of the line between three points be connected with main bifurcation is respectively l
1, l
2, l
3, angle is α
1, α
2, α
3.Definition angle Expressive Features, k
1=α
1/ α
5/ α
9, k
2=α
2/ α
3/ α
4, k
3=α
6/ α
7/ α
8, k
4=α
10/ α
11/ α
12.To l
iand k
ibe normalized,
Then feature interpretation vector is:
S＝[l′
1,l′
2,l′
3,k′
1,k′
2,k′
3,k′
4]
2. Mismatching point rejects the introducing of strategy
In order to improve image registration accuracy further, the present invention adopts the equal criterion of the Euclidean distance ratio between the ratio of the Euclidean distance between the reference picture unique point unique point corresponding with image subject to registration to reject and mismatches on schedule.The precision of registration can be improved so further, mismatch on schedule reject strategic process as follows:
(1) initialization, feature registration point quantity is R
max, between reference picture and image subject to registration, characteristic of correspondence point set is respectively M (i), N (i), makes i=1;
(2) Data (i)=d (M (i), M (i+1))/d (N (i), N (i+1)), wherein d representative is Euclidean distance between 2;
(3) the numerical value h that in Data (i), the frequency of occurrences is the highest is counted;
(4) equal h unique point set M ' in calculating Data (i), N ' is the error hiding set of having rejected.
Beneficial effect of the present invention is:
1. improve the quantity that image characteristic point detects
The present invention can carry out feature point extraction to eye ground image relatively preferably.Accompanying drawing 2 be SIFT method, Harris angle point method, SURF (Speeded Up Robust Features) method and method of the present invention in the feature detection experimental result of different eye ground image, can find out that the present invention is quantitatively better than other three kinds of methods in retinal fundus images feature point detection.Table 1 is the contrast of four kinds of methods extract minutiae quantity in the retinal fundus images that four width are different.
The extract minutiae quantitative comparison of table 1 four kinds of methods
2. improve the degree of accuracy of registration
The present invention can be relatively accurate realization to the registration of medical image.Accompanying drawing 3 be SIFT method, Harris angle point method, SURF method and method of the present invention to the experimental result of eye ground characteristics of image registration, can find out that the present invention is better than other three kinds of methods in the degree of accuracy of retinal fundus images Characteristic points match logarithm and registration.In eye ground image after table 2 is depicted as and chooses translation, rotation and light change four kinds of methods mismatch in registration on schedule with the contrast of registration point group number.
Mismatch in table 2 ocular fundus image registration on schedule with the objective evaluation of registration point group number
In sum, the present invention devises a kind of method for registering being applied to eye ground image.The method introduces bifurcation structure feature interpretation in SIFT method, and registration point quantity and registration accuracy have had larger lifting, can realize the registration of eye ground image well.
Accompanying drawing explanation
Fig. 1 is bifurcation structure schematic diagram;
Fig. 2 is SIFT method, Harris angle point method, SURF method and the method for the present invention feature detection experimental result at eye ground image;
Fig. 3 is the experimental result of SIFT method, Harris angle point method, SURF method and method characteristic registration of the present invention.
Embodiment
Below in conjunction with drawings and the specific embodiments, the present invention is further detailed explanation.
The present invention realizes the registration process of medical image successively through following steps:
Step 1: pre-service is carried out to two images subject to registration, makes its type consistent;
Step 2: extract the SIFT feature structure point set in image, and carry out feature interpretation;
Step 3: extract the bifurcation structure point set in image, and carry out feature interpretation.As shown in Figure 1, a feature structure comprises three neighbor nodes of main bifurcation and its periphery to the method for feature interpretation, k
1=α
1/ α
5/ α
9, k
2=α
2/ α
3/ α
4, k
3=α
6/ α
7/ α
8, k
4=α
10/ α
11/ α
12.To l
iand k
ibe normalized,
Then feature interpretation vector is:
S＝[l′
1,l′
2,l′
3,k′
1,k′
2,k′
3,k′
4]
Step 4: the feature point set obtained according to two width images carries out measuring similarity, determines registration point pair;
Step 5: the registration point pair rejecting mistake, realizes the registration of two width image characteristic points.Wherein, utilize the equal criterion of the Euclidean distance ratio between the ratio of the Euclidean distance between the reference picture unique point unique point corresponding with image subject to registration to reject misregistration point pair.
Finally it should be noted that; above embodiment is only in order to illustrate technical scheme of the present invention and unrestricted; although with reference to preferably embodiment to invention has been detailed description; but protection scope of the present invention is not limited thereto; anyly be familiar with those skilled in the art in the technical scope that the present invention discloses; the amendment that can expect easily or equivalent replacement, and do not depart from the spirit and scope of technical solution of the present invention, all should be encompassed within protection scope of the present invention.
Claims (3)
1. be applied to a method for registering for retinal fundus images, it is characterized in that comprising the following steps:
Step 1: pre-service is carried out to two images subject to registration, makes its type consistent;
Step 2: extract the SIFT feature structure point set in image, and carry out feature interpretation;
Step 3: extract the bifurcation structure point set in image, and carry out feature interpretation;
Step 4: the feature point set obtained according to two width images carries out measuring similarity, determines registration point pair;
Step 5: the registration point pair rejecting mistake, realizes the registration of two width images.
2. a kind of method for registering being applied to retinal fundus images according to claim 1, it is characterized in that: in described step 3, use new unique point and feature interpretation mode, as shown in Figure 1, a feature structure comprises three neighbor nodes of main bifurcation and its periphery, wherein, the length of the line between three points be connected with main bifurcation A is respectively l
1, l
2, l
3, angle is α
1, α
2, α
3, then α
1+ α
5+ α
9=2 π.So k
1=α
1/ α
5/ α
9definite value, in like manner a k
2=α
2/ α
3/ α
4, k
3=α
6/ α
7/ α
8, k
4=α
10/ α
11/ α
12.To l
iand k
ibe normalized,
then feature interpretation vector is:
S＝[l′
1,l′
2,l′
3,k′
1,k′
2,k′
3,k′
4]
3. a kind of method for registering being applied to retinal fundus images according to claim 1, it is characterized in that: in described step 5, introducing mismatches rejecting strategy on schedule and realizes registration, utilizing the equal criterion of Euclidean distance ratio between the ratio of the Euclidean distance between the reference picture unique point unique point corresponding with image subject to registration to reject mismatches on schedule, can improve the precision of registration so further.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510333654.0A CN104933715A (en) | 2015-06-16 | 2015-06-16 | Registration method applied to retina fundus image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510333654.0A CN104933715A (en) | 2015-06-16 | 2015-06-16 | Registration method applied to retina fundus image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104933715A true CN104933715A (en) | 2015-09-23 |
Family
ID=54120869
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510333654.0A Pending CN104933715A (en) | 2015-06-16 | 2015-06-16 | Registration method applied to retina fundus image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104933715A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105761254A (en) * | 2016-02-04 | 2016-07-13 | 浙江工商大学 | Image feature based eyeground image registering method |
| CN106651827A (en) * | 2016-09-09 | 2017-05-10 | 浙江大学 | Fundus image registering method based on SIFT characteristics |
| CN106846293A (en) * | 2016-12-14 | 2017-06-13 | 海纳医信（北京）软件科技有限责任公司 | Image processing method and device |
| CN107358606A (en) * | 2017-05-04 | 2017-11-17 | 深圳硅基智能科技有限公司 | For identifying the artificial neural network and system of diabetic retinopathy |
| CN107564048A (en) * | 2017-09-25 | 2018-01-09 | 南通大学 | Based on bifurcation feature registration method |
| CN108876770A (en) * | 2018-06-01 | 2018-11-23 | 山东师范大学 | A kind of eyeground multispectral image joint method for registering and system |
| CN110728705A (en) * | 2019-09-24 | 2020-01-24 | Oppo广东移动通信有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
| CN111932593A (en) * | 2020-07-21 | 2020-11-13 | 湖南中联重科智能技术有限公司 | Image registration method, system and equipment based on touch screen gesture correction |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080068560A1 (en) * | 2004-12-02 | 2008-03-20 | Knighton Robert W | Enhanced optical coherence tomography for anatomical mapping |
| US20090265287A1 (en) * | 2008-04-11 | 2009-10-22 | Haas Alfred M | Adaptive Programmable Template Matching System |
| CN101593351A (en) * | 2008-05-28 | 2009-12-02 | 中国科学院自动化研究所 | Ocular fundus image registration method based on range conversion and rigid transformation parameters estimation |
| WO2010119596A1 (en) * | 2009-04-16 | 2010-10-21 | キヤノン株式会社 | Medical image processing apparatus and method of controlling same |
| CN102908120A (en) * | 2012-10-09 | 2013-02-06 | 北京大恒图像视觉有限公司 | Eye fundus image registration method, eye fundus image optic disk nerve and vessel measuring method and eye fundus image matching method |
| CN103810709A (en) * | 2014-02-25 | 2014-05-21 | 南京理工大学 | Vessel-based registration method for eye fundus image and SD-OCT projection image |
-
2015
- 2015-06-16 CN CN201510333654.0A patent/CN104933715A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080068560A1 (en) * | 2004-12-02 | 2008-03-20 | Knighton Robert W | Enhanced optical coherence tomography for anatomical mapping |
| US20090265287A1 (en) * | 2008-04-11 | 2009-10-22 | Haas Alfred M | Adaptive Programmable Template Matching System |
| CN101593351A (en) * | 2008-05-28 | 2009-12-02 | 中国科学院自动化研究所 | Ocular fundus image registration method based on range conversion and rigid transformation parameters estimation |
| WO2010119596A1 (en) * | 2009-04-16 | 2010-10-21 | キヤノン株式会社 | Medical image processing apparatus and method of controlling same |
| CN102908120A (en) * | 2012-10-09 | 2013-02-06 | 北京大恒图像视觉有限公司 | Eye fundus image registration method, eye fundus image optic disk nerve and vessel measuring method and eye fundus image matching method |
| CN103810709A (en) * | 2014-02-25 | 2014-05-21 | 南京理工大学 | Vessel-based registration method for eye fundus image and SD-OCT projection image |
Non-Patent Citations (6)
| Title |
|---|
| HOOVER A等: "Locating the optic nerve in a retinal image using the fuzzy convergence of the blood vessels", 《IEEE TRANSACTIONS MEDICAL IMAGING》 * |
| S CHAUDHURI等: "Detection of blood vessels in retinal images matched using two dimensional filters", 《IEEE TRANSACTIONS ON MEDICAL IMAGING》 * |
| S.TAMURA等: "Semiautomatic leakage analyzing system for time series fluorescein ocular fundus angiography", 《PATTERN RECOGNIT》 * |
| 沈奔: "基于局部血管结构的眼底图像配准", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 范雪婷等: "改进仿射尺度不变特征变换算法的图像配准", 《计算机应用》 * |
| 黄琳琳等: "新的眼底图像配准方法的研究", 《计算机工程与应用》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105761254B (en) * | 2016-02-04 | 2019-01-01 | 浙江工商大学 | Ocular fundus image registration method based on characteristics of image |
| CN105761254A (en) * | 2016-02-04 | 2016-07-13 | 浙江工商大学 | Image feature based eyeground image registering method |
| CN106651827A (en) * | 2016-09-09 | 2017-05-10 | 浙江大学 | Fundus image registering method based on SIFT characteristics |
| CN106651827B (en) * | 2016-09-09 | 2019-05-07 | 浙江大学 | A kind of ocular fundus image registration method based on SIFT feature |
| CN106846293B (en) * | 2016-12-14 | 2020-08-07 | 海纳医信（北京）软件科技有限责任公司 | Image processing method and device |
| CN106846293A (en) * | 2016-12-14 | 2017-06-13 | 海纳医信（北京）软件科技有限责任公司 | Image processing method and device |
| CN107358606A (en) * | 2017-05-04 | 2017-11-17 | 深圳硅基智能科技有限公司 | For identifying the artificial neural network and system of diabetic retinopathy |
| CN107564048A (en) * | 2017-09-25 | 2018-01-09 | 南通大学 | Based on bifurcation feature registration method |
| CN107564048B (en) * | 2017-09-25 | 2020-08-21 | 南通大学 | Feature registration method based on bifurcation point |
| CN108876770A (en) * | 2018-06-01 | 2018-11-23 | 山东师范大学 | A kind of eyeground multispectral image joint method for registering and system |
| CN108876770B (en) * | 2018-06-01 | 2021-06-25 | 山东师范大学 | Fundus multispectral image joint registration method and system |
| CN110728705A (en) * | 2019-09-24 | 2020-01-24 | Oppo广东移动通信有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
| CN110728705B (en) * | 2019-09-24 | 2022-07-15 | Oppo广东移动通信有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
| CN111932593A (en) * | 2020-07-21 | 2020-11-13 | 湖南中联重科智能技术有限公司 | Image registration method, system and equipment based on touch screen gesture correction |
| CN111932593B (en) * | 2020-07-21 | 2024-04-09 | 湖南中联重科智能技术有限公司 | Image registration method, system and equipment based on touch screen gesture correction |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104933715A (en) | Registration method applied to retina fundus image | |
| Xiong et al. | An approach to evaluate blurriness in retinal images with vitreous opacity for cataract diagnosis | |
| Masumoto et al. | Deep-learning classifier with an ultrawide-field scanning laser ophthalmoscope detects glaucoma visual field severity | |
| Zhu et al. | Digital image processing for ophthalmology: Detection of the optic nerve head | |
| Kaur et al. | A method of disease detection and segmentation of retinal blood vessels using fuzzy C-means and neutrosophic approach | |
| Khan et al. | CDR based glaucoma detection using fundus images: a review | |
| Chu et al. | Quantifying choriocapillaris flow deficits using global and localized thresholding methods: a correlation study | |
| Nawaldgi | Review of automated glaucoma detection techniques | |
| Suero et al. | Locating the Optic Disc in Retinal Images Using Morphological Techniques. | |
| Sakthivel et al. | An automated detection of glaucoma using histogram features | |
| Rodrigues et al. | Retinal vessel segmentation using parallel grayscale skeletonization algorithm and mathematical morphology | |
| Gour et al. | Challenges for ocular disease identification in the era of artificial intelligence | |
| Zhu et al. | Automated framework for intraretinal cystoid macular edema segmentation in three-dimensional optical coherence tomography images with macular hole | |
| CN108665474B (en) | B-COSFIRE-based retinal vessel segmentation method for fundus image | |
| Agarwal et al. | Automatic imaging method for optic disc segmentation using morphological techniques and active contour fitting | |
| Abdullah et al. | Application of grow cut algorithm for localization and extraction of optic disc in retinal images | |
| Pk et al. | Automated identification of diabetic retinopathy—A survey | |
| Sheet et al. | Cotton-wool spots, red-lesions and hard-exudates distinction using CNN enhancement and transfer learning | |
| Lotankar et al. | Glaucoma Screening using Digital Fundus Image through Optic Disc and Cup Segmentation | |
| Kawde et al. | Early detection of glaucoma disease using image processing | |
| Medhi et al. | Improved analysis of diabetic maculopathy using level set spatial fuzzy clustering | |
| Hutami et al. | Adaptive Localization and Segmentation of Optic Disc using K-means and Active Contour | |
| Siddalingaswamy et al. | Automatic segmentation of blood vessels in colour retinal images using spatial gabor filter and multiscale analysis | |
| Nugroho et al. | Detection of foveal avascular zone in colour retinal fundus images | |
| Sujithra et al. | Automatic detection of glaucoma based ON cup—Disk ratio with polar segmentation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150923 |
|
| WD01 | Invention patent application deemed withdrawn after publication |